Effective Communication for Researchers - Virtual Residency Workshop
Dive into effective communication strategies for researchers in the Virtual Residency Workshop led by Henry Neeman from the University of Oklahoma. Explore research terminology, the mindset gap, and ways to engage with different types of researchers. Discover how to locate and approach researchers for collaborative projects and elevate your communication skills in the academic realm.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. Download presentation by click this link. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
E N D
Presentation Transcript
Effective Effective Communication Communication Henry Neeman, University of Oklahoma Director, OU Supercomputing Center for Education & Research (OSCER) Associate Professor, Gallogly College of Engineering Adjunct Associate Professor, School of Computer Science Virtual Residency Introductory Workshop 2023 Monday June 26 2023
Outline How to Talk to Researchers: Research Terminology Enterprise IT vs Research Computing The Mindset Gap Researcher Types Things to Say to a Researcher How to Find Researchers How to Find Researchers Projects Effective Communication Virtual Residency Workshop, Mon June 26 2023 2
How to Talk to Researchers: Research Terminology
Is Oxygen a Metal? How many of you believe that oxygen is a metal? Effective Communication Virtual Residency Workshop, Mon June 26 2023 4
Oxygen in Real Life Atomic number 8 Chalcogen Key element in life Also fire, rust, water etc http://en.wikipedia.org/wiki/Oxygen Effective Communication Virtual Residency Workshop, Mon June 26 2023 5
Oxygen in Astronomy The universe is made of the following: Hydrogen Atomic number 1 75% of all baryonic mass Most stars are made of hydrogen plasma Helium Atomic number 2 Noble gas (inert) 24% of total elemental mass Other: ~1% (including oxygen AND EVERYTHING ELSE) http://en.wikipedia.org/wiki/Hydrogen http://en.wikipedia.org/wiki/Helium Effective Communication Virtual Residency Workshop, Mon June 26 2023 6
Planets etc What are planets made of? Metals! (In the chemical sense) Cores of iron, nickel etc Earth s core is 89% iron, 6% nickel, 5% other Mantles of silicates http://en.wikipedia.org/wiki/Planets#Mass http://en.wikipedia.org/wiki/Earth Rose Eveleth, Barns Are Painted Red Because of the Physics of Dying Stars. http://www.smithsonianmag.com/smart-news/barns-are-painted-red-because-of-the-physics-of-dying-stars- 58185724/?utm_source=keywee- facebook.com&utm_medium=socialmedia&utm_campaign=keywee&kwp_0=283306&kwp_4=1091891&kwp_1=506963 Effective Communication Virtual Residency Workshop, Mon June 26 2023 7
So Whats a Metal? To a chemist, metals have a very specific chemical definition. H He O https://chem.libretexts.org/Courses/College_of_Marin/Marin%3A_CHEM_114_- _Introductory_Chemistry_(Daubenmire)/04%3A_Atoms_and_Elements/4.6%3A_ Looking_for_Patterns%3A_The_Periodic_Law_and_the_Periodic_Table http://user.astro.columbia.edu/~gbryan/Site/IGM_files/gas_density_z0.png But, to an astronomer (especially a cosmologist), metals are anything that isn t hydrogen or helium. Effective Communication Virtual Residency Workshop, Mon June 26 2023 8
Projection What happens if you put a mathematician, a psychologist and a movie producer into a room and ask them to discuss projection? What does projection mean to a mathematician? What does projection mean to a psychologist? What does projection mean to a movie producer? A key skill that you ll develop by experience: How to tell when people are talking past each other by using the same word to mean different things, or vice versa. Example: Who s on first? comedy routine: https://www.youtube.com/watch?v=sShMA85pv8M Effective Communication Virtual Residency Workshop, Mon June 26 2023 9
What Are Fluids? Colloquial definition: Liquids. Mom s and physician s definition: Something you should drink plenty of when you re sick. https://www.zocdoc.com/answers/9591/does-drinking-fluids-help- when-you-have-a-cold Effective Communication Virtual Residency Workshop, Mon June 26 2023 10
What Are Fluids? (contd) Physical science & engineering definition: Not solids. Computational Fluid Dynamics The most popular fluid studied is air (Earth s atmosphere). [A] substance, as a liquid or gas, that is capable of flowing and that changes its shape at a steady rate when acted upon by a force tending to change its shape. dictionary.com Liquids are incompressible fluids. Social science definitions Something that has room for interpretation (e.g., gender). Describing a feeling, mood or appearance. Finance: An asset convertible into cash. Effective Communication Virtual Residency Workshop, Mon June 26 2023 11
Database Which of these is a database? 1. SQL/Oracle/Access/DB2/MongoDB? OR 2. A large collection of files? Effective Communication Virtual Residency Workshop, Mon June 26 2023 12
Data vs Metadata What s data? What s metadata? Is code data? Effective Communication Virtual Residency Workshop, Mon June 26 2023 13
Scale At quantum scale over femtoseconds, how much does gravity matter? How about at cosmological scale over eons? What are some other disciplines where gravity DOES matter, and some other disciplines where gravity DOESN T matter? For each, EXPLAIN WHY. Effective Communication Virtual Residency Workshop, Mon June 26 2023 14
Science vs Engineering Science is focused on discovery. Engineering is focused on design. In which case: Is a design project research? Do engineers do science research? What is research about software? Effective Communication Virtual Residency Workshop, Mon June 26 2023 15
CS or IT? What happens if a domain scientist refers to CS as IT? Wait, CS people do research? I thought they were just there to help everyone else with their real research ? Effective Communication Virtual Residency Workshop, Mon June 26 2023 16
Is Simulated Data Actually Data? I had a colleague in Chemical Engineering who told me that, if he referred to output from a simulation as data in front of his colleagues, he d be laughed out of the discipline. A compiler designer considers code to be data the stuff that everyone else considers data isn t very interesting. What are some examples of disciplines with very different definitions of data from what we ve already discussed? Effective Communication Virtual Residency Workshop, Mon June 26 2023 17
Do They Know What You Know? NO, RESEARCHERS DON T KNOW WHAT YOU KNOW. The most normal thing in the world is to assume that the person you re talking to knows what you know. The second most normal thing in the world is to be dead wrong about that. Effective Communication Virtual Residency Workshop, Mon June 26 2023 18
Most People Are Bad at This! The probability two people selected at random will share the same concept about penguins is around 12 percent S. Makin, 2023: People Differ Widely in Their Understanding of Even a Simple Concept Such as the Word Penguin. Scientific American. https://www.scientificamerican.com/article/people-differ-widely-in-their-understanding-of-even-a-simple-concept-such-as-the- word-penguin1/ [Researchers] found that only 2 percent of conversations ended at the time both parties desired, and only 30 percent of them finished when [even] one of the pair wanted them to. R. Nuwer, 2021: People Literally Don t Know When to Shut Up or Keep Talking Science Confirms. Scientific American.. https://www.scientificamerican.com/article/people- literally-dont-know-when-to-shut-up-or-keep-talking-science-confirms/ Thanks to Bev Corwin for these references! Effective Communication Virtual Residency Workshop, Mon June 26 2023 19
Exercise 1. Go to: https://docs.google.com/document/d/1NKEZ3xnlciu8YEwM4yiPd5kiwwm7u8shW7NXW62_Iws/edit 2. At the bottom, enter a piece of terminology a term that s used in many (sub-sub-)disciplines and/or colloquially and the name of a research discipline (or sub-discipline or sub-sub-discipline ...) that uses that term, and define what that term means in that (sub-sub-)discipline. Or, you can add a new definition for a new (sub-sub-)discipline to a term that s already listed. This way, we can create a shared glossary of terminology as used in various disciplines. Note that all the terms discussed in this session are already there, so you ll have to add new ones! Effective Communication Virtual Residency Workshop, Mon June 26 2023 20
Enterprise IT Enterprise IT vs vs Research Computing: Research Computing: Natural Enemies, Natural Enemies, or Natural Allies? or Natural Allies?
Enterprise IT & Research Computing Enterprise IT: 5 NINES Secure Established technology Best practices 5 nines: 99.999% uptime = 5 minutes of downtime per year Research Computing: 1 NINES Fast and flexible (turn on a dime) Cutting edge technology (= broken) In some cases, no such thing as best practices 1 nines: 95% uptime = 18 days (438 hours) downtime per year NSF s standard, from NSF solicitations 17-558, 19-587, 20-606: [NSF-funded] production resources should be unavailable as a result of scheduled and unscheduled maintenance no more than 5% of the time. NOTE: OU s supercomputer ~ 99% uptime; OU IT enterprise = 99.995% Effective Communication Virtual Residency Workshop, Mon June 26 2023 22
Enterprise IT Example On Aug 8 2016, Delta Air Lines experienced a power outage in their Atlanta data center that lasted 5 hours. Cost: $150M ($1M for every 2 minutes of downtime) https://money.cnn.com/2016/09/07/technology/delta-computer-outage-cost/ Effective Communication Virtual Residency Workshop, Mon June 26 2023 23
Enterprise vs Research: Incentives Suppose a mission critical capability is needed tomorrow, and the relevant IT system goes down tonight. Tomorrow, what happens to the Enterprise IT people who are accountable for the outage? Therefore, what must Enterprise IT people do to stay in business? Suppose Research Computing isn t on the cutting edge, and thus proposals from the institution are less competitive. Eventually, what will happen to the Research Computing team? Therefore, what must Research Computing people do to stay in business? Effective Communication Virtual Residency Workshop, Mon June 26 2023 24
Enterprise vs Research: How to Resolve? Research Computing can afford to make mistakes: A system that s mostly up but crashes occasionally is fine. 1 24-hour day of HPC downtime = 10-100 lost grad student days 1 grad student = ~$60K/yr fully loaded with fringe+tuition+Indirect => 100 grad student days = ~$16K productivity loss => ~$300-$3000 productivity loss per research group Cost of 5 Nines vs 1 Nines: 5-10x, but budgets are fixed so the actual cost is cutting computing-intensive/data-intensive research productivity by 80-90%. Therefore: Let the research machine go down from time to time, as a tradeoff for having bigger (but less resilient) resources, to maximize research productivity per year, at the cost of occasional lost days. Effective Communication Virtual Residency Workshop, Mon June 26 2023 25
How Much Failure is Normal? In calendar 2021, on OU s supercomputer: ~18M jobs ran; ~486K jobs failed (2.7% of all jobs); of those ~486K failed jobs, ~2800 jobs failed due to server failure (0.6% of failed jobs, 0.02% of all jobs). So: Batch job failure is normal. 16,666,647 COMPLETED 534,175 CANCELLED 261,256 TIMEOUT 21,532 RUNNING 2 DEADLINE 2148 REQUEUED 3437 PENDING 483,648 FAILED 2840 NODE_FAIL Batch job failure due to server failure is insignificant. Effective Communication Virtual Residency Workshop, Mon June 26 2023 26
Implications for IT Research computing is LESS EXPENSIVE than Enterprise IT. But, it s also LESS RESILIENT (1 nines vs 5 nines). So, when a researcher comes to us for help with a specific capability, we should ask: Can this capability tolerate an average of roughly 8 hours of downtime per month? If yes, Research Computing may well be their best bet. If no, Enterprise IT is definitely the right way to go. Effective Communication Virtual Residency Workshop, Mon June 26 2023 27
Research is the Enterprise Testbed Research Computing has only limited best practices. But, technologies currently being adopted by Research Computing are likely to become enterprise requirements in not-too-many years. So, let Enterprise IT watch Research Computing make mistakes, and use those observations to help develop best practices for Enterprise IT. Example: In 2012, OU Research Computing moved to OU s then-new data center in one week, and was the first team in there. That helped prove that the data center was ready for enterprise systems which would have been too risky to move in first. Effective Communication Virtual Residency Workshop, Mon June 26 2023 28
The Mindset Gap The Mindset Gap
The Mindset Gap In the olden days say, 15 years ago we used to say that a typical new Research Computing user came from a Windows or MacOS desktop or laptop background. Those days are long gone . Nowadays, we say that a typical new user comes from an iOS or Android background. How has that changed our job? Effective Communication Virtual Residency Workshop, Mon June 26 2023 30
Mental Distance What s the mental distance between a handheld vs Linux, command line, remote, shared, batch computing? Installing software Handheld: Tap 3 times. Large scale: EasyBuild or Spack if you re lucky, configure/make with lots of dependencies if you re unlucky, bizarre random weirdness in practice. Is it realistic to expect all of our users to be able to do this? Installing storage Handheld: Buy a card for $10-50, pop it into the slot, the OS automatically recognizes it and starts using it. Large scale: RFP, bid evaluation, configuration, purchase, deployment, maintenance, expansion, decommissioning. Effective Communication Virtual Residency Workshop, Mon June 26 2023 31
Whats the Cost of Storage? Handheld: tens or hundreds of dollars (which gets you tens or hundreds of GB). Laptop: tens or hundreds of dollars (which gets you TB of spinning disk or GB/TB of SSD). Large scale (per copy) ~1 PB usable tape: ~$6K ~1 PB usable spinning disk : ~$47K (ultra-cheap version) => ~8x ~1 PB usable SSD: ~$329K (ultra-cheap version) => ~56x And, for large scale resources, PB are normal! Effective Communication Virtual Residency Workshop, Mon June 26 2023 32
Why Else is Using Research Computing Hard? Research Computing is hard because: 1. the mindset gap; 2. Research Computing systems always get harder to use: the storage hierarchy get deeper (e.g., registers/cache/RAM/Optane/flash/spinning disk/tape), parallelism gets more hybrid (e.g., MPI on top of GPUs); 3. more researchers need Research Computing: more disciplines need Research Computing (e.g., life sciences, social sciences, humanities, arts, business, athletics), a higher fraction of users within each discipline need research computing. Effective Communication Virtual Residency Workshop, Mon June 26 2023 33
Why Are Researchers Why Are Researchers That Way ? That Way ?
Researcher Types Faculty Tenure-Track Faculty Tenured Faculty Research Faculty Staff Postdocs Students Effective Communication Virtual Residency Workshop, Mon June 26 2023 35
What Are Faculty Rewarded For? Faculty at research-intensive institutions are rewarded for three things: bringing in grant money; publishing papers; graduating students. Faculty absolutely AREN T rewarded for having good IT. So they d strongly prefer NOT to pay more than the ABSOLUTE MINIMUM for their computing (ideally zero). (And their mental model for what compute and storage cost is the price of a laptop and some USB hard drives from their local big box store because that s been their experience so far.) Effective Communication Virtual Residency Workshop, Mon June 26 2023 36
Tenure-Track (Not Yet Tenured) Faculty At research-intensive institutions: Incentive Structure: I need to (a) publish lots of papers, (b) bring in lots of grant money and (c) graduate lots of students, or I m fired. Need: I need stuff to work now and keep working reliably. Timeline I have 7 years (typical tenure-track duration), BUT I have 6 years (the 7th year is finding a job elsewhere if I don t get tenure here), BUT I have 5 years (the 6th year is when my materials are evaluated), BUT I have 4 years, because it typically takes a journal article about 6 months from submitting it to it getting published. Effective Communication Virtual Residency Workshop, Mon June 26 2023 37
Tenured Faculty At research-intensive institutions: Incentive Structure: I need to publish lots of papers, bring in lots of grant money and graduate lots of students, or else: I won t get a raise; I won t get a named chair; I won t get other prestigious outcomes (e.g., elected to the relevant National Academy etc). Need: I need stuff to work now and keep working reliably. Effective Communication Virtual Residency Workshop, Mon June 26 2023 38
Research Faculty (Non-Tenure-Track) If I don t bring in grant money, I m laid off. I need to publish a lot to keep bringing in grant money. I need a track record of graduating lots of students, so I can get a tenure track job somewhere. Because I don t want to have to live on soft money forever! Effective Communication Virtual Residency Workshop, Mon June 26 2023 39
Postdocs I need to publish a lot or I ll lose my postdoc position. I need to learn how to get lots of grant money, and actually get some of my own, so I can get a permanent position. Effective Communication Virtual Residency Workshop, Mon June 26 2023 40
Students My first goal is to graduate. Anything that delays graduation costs me money: I may or may not have an assistantship, scholarship, grant, etc. While I m in school, I m giving up that many years of salary and benefits. Effective Communication Virtual Residency Workshop, Mon June 26 2023 41
Research Funding Research Funding
Academic Research Funding Model Most academic research is funded by grants: a random amount of money shows up on a random date and is available for a random duration. Grants usually strongly favor researcher salaries and disfavor equipment, services and non-research staff. Researchers don t get much credit for the latter. Recurring charges are very hard to pay long term. Some academic research at some institutions is internally funded (for example, via endowment), but this is typically larger and more common at higher-ranked institutions. The problem with the Top 50 is, there s only 50 of them. Effective Communication Virtual Residency Workshop, Mon June 26 2023 43
Probability of Success #1 National Science Foundation, federal FY2021: 26% overall Engineering (ENG): 20% Technology, Innovation & Partnerships (TIP): 20% Social, Behavioral & Economic (SBE): 23% Computer & Information Science & Engineering (CISE): 24% Office of Advanced Cyberinfrastructure (OAC): 38% Education & Human Resources (EHR): 20% Office of the Director: 29% Biosciences (BIO): 30% Mathematical & Physical Sciences (MPS): 30% Geosciences (GEO): 45% Funding is governed by the Law of Large Numbers: You have to submit lots of proposals to get any funding. So faculty can expect to spend a lot of time writing proposals. http://dellweb.bfa.nsf.gov/awdfr3/default.asp Effective Communication Virtual Residency Workshop, Mon June 26 2023 44
Probability of Success #2 National Science Foundation, federal FY2021: 26% overall EPSCoR jurisdictions (mean per state 24.4%): GU 0%, VI 14%, KS/MS/SD 19%, SC 20%, AL/NE 21%, IA/ND 22%, DE/KY/OK/WV 24%, ID/NV/NH 25%, LA 26%, NM/PR/VT 27%, AR/ME/WY 28%, HI 32%, MT 35%, RI 36%, AK 42% Non-EPSCoR jurisdictions (mean per state 26.5%): FL 20%, GA/MO/OH/TN 22%, TX 23%, VA 24%, AZ/MI 25%, CT/IL/IN/MN/UT 26%, MD/NY/NC/PA 27%, CA/MA/NJ 28%, WI 29%, CO 31%, OR 33%, WA 34%, DC 35% Funding is governed by the Law of Large Numbers: You have to submit lots of proposals to get any funding. So faculty can expect to spend a lot of time writing proposals. http://dellweb.bfa.nsf.gov/awdfr3/default.asp Effective Communication Virtual Residency Workshop, Mon June 26 2023 45
Research Computing Funding Models Most academic institutions fund research computing via a combination of some or all of: internal funds on a regular cadence (annual budget); internal funds at random times (Let s buy a supercomputer!); grant funds at random times (Let s buy a supercomputer!); for public institutions, state appropriations in random years (the legislature giveth and the legislature taketh away); usage charges (uncommon); condominium purchases of researcher-owned compute servers at random times. Also, some institutions like leasing, others like buying, others do both. Effective Communication Virtual Residency Workshop, Mon June 26 2023 46
How Should Faculty Spend Their Time? At OU, we don t ask faculty to write grant proposals for time on our supercomputer. Why? Faculty have a limited number of hours per year for writing proposals. We d much rather they spend that time writing proposals for external research grants, than for internal time on a machine that OU has already paid for. Effective Communication Virtual Residency Workshop, Mon June 26 2023 47
Things to Say to a Things to Say to a Researcher Researcher
Cost This other way of doing it is cheaper than how you re currently doing it. For the same cost, it could be so much better. Effective Communication Virtual Residency Workshop, Mon June 26 2023 49
Control You get to decide how to use your piece. You can share it with whoever you want. Effective Communication Virtual Residency Workshop, Mon June 26 2023 50